Semiconductor

What is semiconductor?

Semiconductor is material having nature intermediate between conductor and insulator. Current can easily flow through conductor, while it cannot through insulator. Although semiconductor is not insulator, little current flows through its element.
Conductor such as metals contains many free electrons, and these electrons easily move. As a result, current flows through conductor.



At first, germanium (Ge) was used as the element of semiconductor. At present, silicon (Si) is mainly used.



Usually, the expression of a semiconductor device means what is made from semiconductor. For example, pure silicon hardly conducts electricity, but if it is mixed with other element (by means of ion implantation with high-temperature diffusion: doping), the resultant mixture could conduct electricity. By doping elements with various characteristics to silicon, various operations can be realized. For the production process of semiconductor devices, refer to "How Semiconductors Are Made" on NEC Electronics' website.

An atom of silicon (Si) has four electrons (negative charge) on the outermost shell. If a silicon atom has eight electrons on its outermost shell, it becomes stabilized. If each silicon atom is covalently bound by four other silicon atoms, therefore, a stable silicon crystal is created (that does not conduct electricity).





A silicon atom has four electrons (hands for bonding) on the outermost shell. If, for example, indium (In), which has three electrons (hands for bonding), is doped to silicon, it is bonded to silicon and, as a result, it runs short of one electron. This electron deficiency status is called a hole, into which an electron from another atom comes easily. In the next time, as result to have given an electron to the hole, the atom has a hole. Electrons can move between holes like this as free electrons. This nature is called positive (P type).
If phosphor (P), which has five electrons on the outermost shell, is doped to silicon, one electron is left as remainder of bonding. This electron is called a free electron. The free electron is easy to move to other atoms. This nature is called negative (N type).
A semiconductor device consists of P type and N type semiconductors.
Free electrons (negative charge) and holes (positive charge) are called carriers.

What is the operating principle of a semiconductor device?

When a P-type semiconductor and an N-type semiconductor are bound, the bound portion becomes a stable area called a depletion layer.
If a battery is connected, with the positive pole to the P-type side and negative pole to the N-type side of these joined semiconductors, the depletion layer narrows, carriers can move between the two semiconductors and current can flow between them. Conversely, if the negative pole of the battery is connected to the P-type side and the positive side to the N-type side, the depletion layer widens, and carriers can no longer move between the two semiconductors, and consequently, no current can flow.
This is rectification and the operating principle of a diode. The name diode means di-electrodes (two electrodes).



If a voltage so high as to break the depletion layer is applied in the reverse direction, a current can flow through in the reverse direction. This voltage is called a breakdown voltage and is constant (stable) regardless of the current. A diode that uses the breakdown voltage as a constant voltage is Zener diode. Therefore, Zener diode is connected in the reverse direction to the ground, so that it absorbs noise voltage higher than a specific voltage on a signal line.



If three semiconductors, say, two P types and one N type or two N types and one P type, are alternately joined, like two diodes joined face-to-face or back-to-back, no current flows between the both ends, called collector (C) and emitter (E), of the junction. This is the structure of a transistor. Transistor is named for "transfer resister". In addition, because both holes and electrons are used as carriers, this transistor is called a bipolar transistor. Bipolar means having two (bi) polarities.



The layer called base (B) in the middle is very thin. If a charge having a reverse polarity to the carrier is applied to the base, current easily flows between the collector and emitter. A voltage that is applied to make the transistor active is called a bias.
Changes in the base current I_B bring about substantial changes in the collector current I_C that flows between the collector and emitter. This is what a bipolar transistor amplifies current. Current amplification factor (I_C/I_B) when the emitter is commonly grounded is expressed as h_FE (FE: Forward, grounded Emitter). (β was used in the past but, today, h is uniformly used as one of h parameters.)



Although an example of a junction transistor is shown above, the current mainstream is the planar type transistor, such as an IC, with P type and N type created by repeating diffusion on a silicon substrate.



As opposed to a bipolar transistor, there is also a unipolar transistor (having one pole). This transistor uses either holes or electrons as carriers. The transistor of this type is called a field effect transistor (FET) and comes in junction type (J-FET) or insulation type (MIS).
In a junction FET, the size of the depletion layer changes depending on the electric field generated by gate G voltage, and the width of a channel (path) changes. As a result, current between both the ends, source S and drain D, is controlled. This type of transistor is suitable for power application, and is not suitable for integrated circuits because of its structure.



In an insulation FET, the gate electrode is not directly connected to a semiconductor. Instead, it is placed on an oxide film (insulation film) over the semiconductor. As a result, no current flows through gate G and between S and D because the channel facing the gate serves as a depletion layer. When an electric field is generated by a gate potential on the oxide film and thus charges are accumulated on the channel, current flows between S and D.
This structure is called metal insulator semiconductor (MIS) in which an insulator is placed between electrodes and semiconductor, and especially a transistor that has an oxide film as an insulator is called a metal oxide semiconductor (MOS). In other words, a transistor of this type is generally called a MOS FET or MOS transistor. Depending on the polarity by which the channel conducts, it is also called P channel MOS (PMOS) or N channel MOS (NMOS). Complementary MOS (CMOS) transistor is a transistor that consists of PMOS and CMOS in combination.





The gap p-p or n-n of the channel is called a channel length that is used as a scale of density. The length of the width of the channel is called a channel width.

There are both high-speed and low-speed semiconductor devices, but how do they differ?

The operation speed of a semiconductor device is determined by the signal (electron) transmission time. Time is inversely proportional to speed and proportional to distance.
Given the same distance, a faster speed means a shorter transmission time. The transmission speed differs depending on physical characteristics. Semiconductor devices consist of P type and N type parts, and the mobility of electrons depends on the elements that are included in these parts and their density, resulting in the difference of speed. Naturally, mobility also depends on the specific resistance of the metal used in wiring and pins, but not to the point of affecting the operation speed of a given device.



On the other hand, given the same transmission speed, the shorter the distance, the shorter the transmission time. Therefore, as process technology (expressed for example as "0.35 µm process," where "0.35 µm" is called the design rule) becomes increasingly miniaturized, transmission distances are becoming shorter. The design rule indicates the channel length (in CMOS, the interval of the P channel or the N channel) or the minimum wire pitch, and as these become smaller, the total chip size shrinks. By this, operation speed is becoming faster and operation voltage is becoming lower.
However, recent processes have entered under 100 nm (one 10,000,000th of a meter) level, which involves large capacitance and resistance of wire, making it more difficult to achieve high speeds simply through process miniaturization.



Furthermore, operation speed also depends on the circuit configuration. The larger the number of element stages, the greater the delays and thus the slower the speed. Particularly in the case of clock synchronization, the next operation being started only at a new clock cycle, the operation speed is limited by the number of element stages operating in a clock cycle.



The operation speed of semiconductor devices depends on the product, based on the combination of the above-described factors.

(2007/11)

What is the difference between an IC and an LSI?

The only difference is the scale. The names IC and LSI do not express differences in function or performance.
IC stands for Integrated Circuit, and LSI, for Large Scale Integrated circuit.
In the 1970s when ICs increasingly became large in scale, classification such as into SSI (Small Scale Integrated circuit), MSI (Medium Scale Integrated circuit), LSI, VLSI (Very Large Scale Integrated circuit), and ULSI (Ultra Large Scale Integrated circuit) was used. But expression could no longer catch up with a progress in integration. Today, therefore, the name IC is generally used and especially only IC that integrates more than several 1000 transistors is called LSI.
A single transistor or diode that is not integrated is called a discrete component.

What are the advantages of increasing the integration of ICs?

First, looking at the system as a whole, the number of parts is reduced, which increases reliability. Moreover, the parts that must be replaced in case of a fault can be identified more easily, facilitating maintenance. Furthermore, in the case of custom products, circuit analysis is more difficult, resulting in higher confidentiality.
Next, in the case of miniaturizing process technology, in the area of characteristics, higher speed can be expected. Moreover, in this case, operating voltage tends to be lower, which allows the realization of lowering power consumption and heat generation (and therefore CO₂ reduction). Further, as chip sizes become smaller, the chip can be fitted into smaller packages allowing reductions in mounting surface and contributing to system miniaturization.
However, recent processes have entered the under 100 nm (one 10,000,000th of a meter) level, which involves relatively large capacitance and resistance of wire, and further increases in speed through the use of miniaturizing processes are becoming difficult to achieve. Moreover, with miniaturization, the off-leak current during standby increases, so that lower power consumption simply though process miniaturization are difficult to achieve. NEC Electronics has successfully achieved large reductions in power consumption through the use of a high-K gate insulation film and the like.

(2007/11)

How do the number of transistors and the number of gates indicating the density differ?

The number of transistors indicates the total number of internal bipolar transistors and MOS transistors (FETs). In contrast, the number of gates indicates the number of MOS gate structures (this does not mean the number of gate pins of each FET). Basically, a gate for a CMOS device consists of two MOS transistors (NMOS and PMOS). Therefore, the number of gates is half the number of transistors. However, one transistor (NMOS only) creates a gate structure for an open-drain output device. This is counted as a difference in the number of gates from those of the CMOS device.
The "number of usable gates" of a gated array means the number of CMOS structures that can be used as a logic gate of a circuit which users can design.

(2008/04)

Why is a CMOS device easily affected by static electricity?

A bipolar device is resistive and absorbs static electricity because current flows with base input transmitting through the semiconductor.
In contrast, a CMOS device has gate input insulated by an oxide film. Consequently, no current flows and the voltage of static electricity is applied to a thin oxide film as is. For this reason, there is a strong possibility that the oxide film breaks down.

What are the different types of semiconductor products, and how are they used?

The main semiconductor products are classified by function as follows.
At NEC Electronics, we call communication devices and display/image processing devices "ASSP" (Application Specific Standard Product), and distinguish them from semi-custom ASIC (Application Specific IC).



(1) Diode
A diode is a rectifying device that lets the current flow only in one direction. However, if the rated voltage (breakdown voltage) is exceeded even in the opposite direction, the device changes to a conductive state and current flows through. For the operation of diodes, refer to the operating principles of semiconductor devices.
(a) Rectifying diode
A rectifying diode uses a rectifying function to obtain a direct current from an alternating current power supply. Here, in the case of one diode, for half-wave rectification not output on the minus side, a pulsating flow is output, and even if a smoothing capacitor is installed, the DC voltage fed to the load becomes lower.
In a bridge rectifying circuit that uses four diodes, two pairs each consisting of two facing diodes are combined, neighboring diodes make the plus side and minus side of AC flow in the forward direction on an alternating basis, the voltage supplied to the load is continuously full-wave rectified, and a DC voltage is obtained by using the smoothing capacitor. Moreover, even in the case of a two-phase full-wave rectification circuit in which the center tap of the transformer is grounded, the same DC voltage is obtained.



(b) Switching diode
A switching diode uses a rectifying function as a small-signal high-speed switching element, rather than passing a large current such as power supply.
(c) Zener diode
Using the fact that the device enters a conductive state when the breakdown voltage at the reverse polarity is exceeded, the Zener diode is used for circuits that require a constant voltage, as well as in circuits where it prevents destruction from overvoltage, surges, or static electricity. When used for noise elimination, it is sometimes called a noise clipping diode from this application, and when it is used mainly to absorb surges and static electricity, it is also sometimes called a surge absorber.
During regular operation without noise clipping, the capacitance between pins affects the time constant of signal changes. Therefore, when used for high-speed signals, a small inter-pin capacitance is required.



(2) Transistor
A transistor is an element that performs amplification or switching operation depending on input signal changes. For the operation of transistors, refer to the operating principles of semiconductor devices.
(a) Bipolar transistor
Bipolar transistors are used mainly for current amplification.
Naming the current amplification rate (maximum change ratio of output/input) at emitter ground "h_FE", h_FE ranks are specified for each product. Bipolar transistors are classified by polarity into PNP and NPN types, and by characteristic into types for low frequencies and high frequencies.
Bipolar transistors include small signal transistors and power transistors for power applications. Moreover, some power transistors boost the amplification rate with a Darlington connection.
Transistor arrays can also be formed by aligning multiple bipolar transistors.
(b) FET (Field Effect Transistor)
A FET is a field effect transistor that lets only a small current flow into input.
A JFET (junction FET) is a transistor that is used for amplitude or impedance conversion through current control by applying a backward bias between the gate and the source. This is a depletion-type transistor that always allows current to flow between the drain and the source, even when no voltage is applied to the gate. The drain current is controlled by the gate voltage.
MOS FETs are used mainly for circuit switching. They are classified as P-channel and N-channel MOS FETs, based on polarity. They are enhancement-type transistors which make current flow between a drain and a source only when a voltage is applied to a gate.

(3) Thyristor
A diode consists of a PN junction, and a thyristor has a PNPN configuration achieved by placing two such diodes in series. The name "thyristor" comes from Thyratron Transistor, which is a transistor that performs thyratron operation combining PNP and NPN transistors. It is a switching element that functions as a diode by passing a constant current to the P pole between the N poles, and using the P pole as a gate. A standalone thyristor is called a silicon controlled rectifier (SCR).
The combination of two SCRs used for DC control is called a triac.



(4) Microcomputer
A microcomputer is a semiconductor device that incorporates a CPU (Central Processing Unit), which is characterized by the fact that its operation can be defined freely through program processing. The basic function of a microcomputer is input/output, transfer, and calculations. Calculations are executed with arithmetic operations (addition, subtraction, etc.) and logical operations (AND, OR, etc.), using the ALU (Arithmetic Logic Unit) in the CPU. Input/output and transfer targets are peripheral devices and memory. By contrast to the MPU (Micro Processing Unit), which is operations performed in a CPU main, recently MCUs (Micro Control Unit) that incorporate the peripheral units and memories in a chip have become widely used for embedded applications.



The CPU involves a concept called architecture, which defines various aspects including data bit width and instruction sets, address space, addressing mode, peripheral resources allocation. Some of the data bit widths currently employed in microcontrollers are 4, 8, 16, 32, and 64. Giving the data bit width a crosswise orientation, addresses have a lengthwise orientation, and their product is the amount of data that can be controlled. Two methods can be used, one whereby memory and I/Os are controlled as separate spaces, and the other whereby I/Os and memory are allocated in the same space (memory mapped I/O).



Peripheral devices include I/O devices, that perform inputs/outputs, such as display output, key input, or disk device input/output, and support devices, which perform timer and interrupt control, DMA control, etc.
Memory consists of ROM and RAM. Normally, programs are written to ROM, but in some cases, they may be written to RAM for rewriting purposes. In personal computers, just the startup program is located in ROM, while the OS and application programs are read out from the hard disk, a CD-ROM, etc., and are deployed in RAM to be executed. The program area may be in internal memory or external memory. Mask ROM is currently giving way to flash memory as the most widely used type of internal ROM.
The CPU first reads out instruction code from the memory (operation code fetch), analyzes it with an instruction decoder, and decides the operation to be executed. At this time, the program counter (PC) indicates the memory execution address for that instruction code. It is updated at each program execution, so that it always indicates the next address. In recent microcomputers, rather than executing the next operation code fetch after each instruction execution, some instructions are loaded one after the next to a prefetch queue, and also these instructions are continuously executed in pipeline, resulting in raising execution efficiency.
In the case of data input/output and operations using an ALU, an accumulator (not mounted in recent microcomputers) is used as a buffer that temporarily holds the data. In the case of data transfer to/from memory and input/output to/from I/Os, the target address is output to the address bus, and it is decoded by an address decoder, the target device is selected, and data transfers with the accumulator are done via the data bus.
In the case of recent microcomputers that process large amounts of data, frequently accessed data is placed in internal cache memory for higher access efficiency.
In the case of operations, the two values to be used are input to the ALU from the accumulator and a general-purpose register and calculated, and the result is input to the accumulator.
In the case of interrupts and sub-routine calls, branching is executed after saving the current address (PC content) as the return destination address to the memory (RAM) stack area, but that saving address is managed by the stack pointer (SP), and during save/recovery, these contents (addresses) are output to RAM via the address bus.

(5) Memory
This is a semiconductor device that stores data in a microcomputer system, etc.
(a) ROM (Read Only Memory)
This is a memory that can be read only and that retains its data even the power supply is switched off (nonvolatility). It is used to store fixed programs, data tables, etc. There are several types of ROM, including mask ROM, which is written to through a masking process during fabrication, and PROM (Programmable ROM), which can be written electrically by the customer after fabrication. PROM is divided into two types, EPROM (Erasable PROM), which can be rewritten, and OTP (One Time PROM), which cannot be rewritten (it can be written only once). Formerly, all EPROMs were UVEPROM, which was written after being erased with UV rays, but now EEPROM, which can be erased and rewritten electrically, is commonly employed. EPROM is used for prototypes when programs have not yet been completed.



(b) RAM (Random Access Memory)
This is a memory that can be freely written and read during system operation, and which loses its written data when the power supply is switched off (volatility). It is used to temporarily save programs and data. In case of data rewrite, it is overwritten.
RAM comes in two types, SRAM (Static RAM), which has a cell structure that uses flip-flop circuits to hold data as long as the power supply is on, and DRAM (Dynamic RAM), which has a capacitor-type cell structure that loses its data even if power is supplied, unless a refresh signal is periodically supplied. SRAM is easy to use and has excellent speed characteristics. DRAM can support large capacities, but requires refresh control. The cell addresses of SRAM are assigned linearly using address lines A0 to An. The cell addresses of DRAM are assigned in a matrix using row addresses A0 to Am and column addresses A0 to An, which are activated by the RAS (Row Address Strobe) and CAS (Column Address Strobe) signals, respectively. When the RAS signal becomes active, the electric charge in the memory cell connected to the selected word line is passed to the bit line, the signal difference is amplified with a sense amplifier, and the cell that was charged is charged again. This is the DRAM refresh operation.



(c) Flash memory
This is a type of EEPROM memory, but instead of rewriting 1 byte at a time, erase and write is done in block units. Formerly there were two types of power supplies with also the power supply pin (Vpp) for write, but more recently rewrite on the system has become possible with just one power supply, and thus this memory is now positioned between ROM and RAM, and tends to be treated as a separate category.
The latest microcontrollers (MCUs) of NEC Electronics generally incorporate flash memory as featured by the company's All Flash policy. For the internal memory of microcontrollers, refer to the type of internal ROM of microcontrollers.

(6) Peripheral devices
Peripheral devices are devices located around the CPU that perform signal input/output operations and data processing that cannot be done by the CPU on its own.
(a) Communication devices
Communication devices are devices that are placed as bidirectional gateways between the transmitting side and the receiving side for data transfers between systems. Mutual transmission and reception can be performed by matching the transmission speed and protocol (communication procedure and format) of both sides.
Communication is largely divided into trunk system communication such as ATM (Asynchronous Transfer Mode), and interfacing between various devices such as USB. By the other classification, based on data alignment, there is parallel communication for transmitting and receiving like data bus image, and serial communication, whereby serial data is converted (serialized) on the transmitting side, and is then reconverted (deserialize) into parallel data on the receiving side. In the case of parallel communication, the number of lines required is the same as the number of bits, while in serial communication, a conversion circuit (SERDES: Serializer/Deserializer) is required. In a high-speed interface example, there is serial PCI Express for parallel PCI.
In serial communication, communication time corresponding to the bit array is required, and in order to accurately executed transmission/reception until the last bit, the synchronous method using the same clock as reference for the transmitting side and receiving side is frequently employed. Bit synchronization methods such as UART (Universal Asynchronous Receiver Transmitter) are also commonly employed, but such use is limited to cases under several hundred kbps and with a speed error of 2% or less.
Moreover, communication devices generally also incorporate several FIFO stages as transmit buffers and receive buffers, so that the transmit data can be written in rapid succession by the CPU on the transmit side, and so that the receive data can be received in rapid succession. By the CPU on the receiving side even if the CPU processing is slow.



Full duplex, which allows simultaneous communication in both directions, and half duplex, which switches the communication direction, are available as bidirectional communication methods.
(b) Display and image processing devices
A display device outputting special information such as characters is an IC for OSD (On-Screen Display) and is called DSD solely. OSD is used for example to display remote controller operation menus on a TV screen.
Image processing devices output image data, process shape, color, and other changes, or perform MPEG (Moving Picture Experts Group) encoding/decoding (CODEC: coder/decoder) and compression/expansion which allows data volume reduction for transfer and saving. Programs and character data give completely different results if even just one bit is missing, so that lossless compression is required, but in the case of image and audio data, small amounts of missing data go undetected by humans, so that lossy compression is widely used for such types of data to obtain large compression rate.
NEC Electronics offers various types of digital AV LSIs, including the EMMA (Enhanced Multimedia Architecture) Series, which incorporates an MPEG codec.
CCD sensors (image sensors) are another type of device, the image input device. A CCD, which stands for Charge Coupled Device, accumulates electric charges from the light received by photo diodes, and transfers the charges as image data.
Display formats include the bit matrix format, which divides images into dots placed to a matrix pattern, and the segment format, which combines fixed forms like digits of a digital clock. In the dot matrix format, increasing the division number results in smaller dots, smoother shapes, more picturesque gradations, and obtaining highly detailed images. However, the amount of data greatly increases, which makes compression/expansion important. Moreover, in the case of color, each dot is broken down into the three primary colors (RGB: Red, Green, Blue), and each data is processed as a pixel. By using grayscale expression (for example, 0 to 255 for 8 bits) to adjust the brightness of each pixel data, the hue and tone of each dot can be adjusted.

[Tea Break] The combination of the three primary optical colors creates white (no color). However, the three primary colors for paint are red, blue, and yellow, and the combination of these three colors yields black. Black in optics is the absence of either of these colors. Since ink is used on the paper output by a printer, color data is converted, using cyan (C), magenta (M), and yellow (Y), as well as black (K) to represent a deep black. In other words, different color models are used for different situations, such as RGB for displays and CMYK for printing.

(c) A/D converter, D/A converter
A/D converters (ADC) convert analog signals into digital signals. By converting the outputs of various sensors via an A/D converter, analog data can be processed and transmitted, and saved as digital data. The other way around, D/A converters (DAC) convert digital signals into analog signals. For example, by combining these two types of converters, it is possible to transfer, record, and replay digital broadcasts and DVD audio and video. Recently, these converters tend to be built into microcontrollers and ASSPs.
A/D converters sample the input voltage at fixed intervals, and numerically convert the sampled values into values expressed as levels into which a full scale is divided. For example, in the case of a 12-bit A/D converter using the reference voltage as the full-scale voltage, that full scale is divided into 4096 levels, and the conversion result from n/4096 is n (000H to FFFH).



In a D/A converter, digital data is sequentially converted into analog voltages. If the output interval is made to match the sampling cycle in the A/D converter, an analog waveform can be generated.

(7) Operational amplifier, comparator
An operational amplifier, also called an OP amp, is a circuit that executes differential amplification of analog input and reference voltage, and is used for amplification and analog computations (addition, subtraction, differential calculus, integral calculus). Using the computation functions, the OP amp can be used also in the input stage of the A/D converter. For details, refer to the OP amp FAQ.
These are two types of operational amplifiers, the voltage differential type, and the current differential type using an emitter ground transistor for input (Norton amplifier). NEC Electronics manufactures the voltage differential type.
Comparators are circuits that compare the input signal with reference voltage. They are used for both analog input and digital input, and the comparison results are output as digital data. For details, refer to the comparator FAQ.

(8) Power supply IC
Automotive batteries and in-house power lines cannot be used for power supplies for electronic circuits because their voltage is too high or they supply AC power. Therefore, it is necessary to perform voltage conversion or rectification (refer to (1) Diode)). At this time, ripple elimination and DC-DC conversion is done by the power supply IC. Power supply ICs are used for example in the internal power supply of electronic equipment and AC adapters. For details, refer to FAQs on the power supply IC.

(9) Standard logic
This is the basic logic such as inverters and logic ICs (AND, OR, etc.). At present, standard logic is provided as user logic in the form of function blocks of gate arrays, etc.

(10) ASIC
This is a type of IC in which the standard circuit elements are placed on low layers. Surface wiring is formed with a product mask depending on the customer's system circuit to complete the product. Development time reductions, security, and mass production cost reductions can be realized by using ASICs. Since ASICs are positioned between standard products and full-custom products, ASICs are classified as semi-custom products.
(a) Gate array
This is a digital ASIC configured of MOS gates arranged in a grid array, and standard logic such as AND and OR and combination circuits thereof are configured by wiring.
(b) Cell-based IC
This is an ASIC that provides large-scale macros such as CPUs and communication units as IP cores. Analog macros such as A/D converters and D/A converters are also available.
(c) Analog master
This is an analog ASIC that incorporates bipolar transistors, resistors, and capacitors. Other elements that are incorporated in analog masters are operational amplifiers and comparators.
(d) Mixed signal ASIC
This is a digital/analog mixed ASIC that incorporates gate arrays and analog master functions.

(2007/10)

What is a compound semiconductor?

Semiconductors can consist of a single element (for example, silicon (Si)), or of a crystal formed artificially from two or more elements (for example gallium (Ga) and arsenic (As), which have a long track record of being used for semiconductor products).
The latter is generally called as compound semiconductor.
The above-mentioned gallium and arsenic compound (III-V Family) is characterized by an electron mobility rate that is approximately six times greater than that of silicon, it is used as the substrate material for high-frequency/high-speed semiconductor devices.
Other compound-use elements (Al: Aluminum, In: Indium, N: Nitrogen, P: Phosphorus, etc.) are used as substrate materials for light-emitting diodes (LDE) ^(Note).
Zinc (Zn) and Tellurium (Te) are commonly used as a p-type impurity and n-type impurity, respectively, for III-V Family compounds.

Note:
When a forward current flows through a light-emitting diode, the energy from the recombination of those positive holes and electrons is smaller than the original energy of the positive holes and electrons, and the difference of energy is emitted as light. Regarding such light emission, in the case of a single element, light emission is difficult, while in the case of a compound, light is readily emitted, so that compounds are used as the substrate of light emitting diodes.
Since the wavelength of the emitted light depends on the elements of the compound used to make the diode, the color of the light can be changed by changing the elements.

(2007/10)

What is an optical semiconductor?

The laser diodes used in the familiar DVD player convert electric signals into optical signals. Conversely, the CCDs (Charge Coupled Devices) used for example in digital cameras convert optical signals into electric signals.
Semiconductors that have the capability of converting electric signals into optical signals or vice-versa are called optical semiconductors.

(2007/10)